A Simple Data-Parameters Balancing Framework for Early Ventricular Activation Origin Localization
Keywords: Early Ventricular Activation Origin Localization;
Abstract: Accurately identifying the site of origin (SoO) of early ventricular activation is crucial for catheter ablation, an effective therapeutic option for treating ventricular arrhythmia. However, due to the limited availability of clinical data and the errors introduced during data preprocessing, achieving precise localization remains a challenge. While deep learning models offer an end-to-end approach for data input in the ECG field, they often suffer from overfitting caused by limited training data, hindering continuous performance improvement. This paper proposes a Simple data-parameters Balancing framework for early ventricular activation Origin Localization (SimBOL). By using onset-based data augmentation, the SimBOL method expands the training data derived from clinical samples. The framework utilizes a small-scale 1D convolution model that balances the relationship between available training data and model complexity, effectively mitigating overfitting and eliminating the need for extensive data preprocessing.SimBOL achieves a localization error as low as 9.83 mm, which meets clinical acceptable localization error < 10 mm and outperforming existing methods in predicting the SoO of early ventricular activation. The discussion about data augmentation and model architecture on ECG signal processing, offering new insights into optimizing deep learning applications for ECG-based tasks.
Primary Area: applications to physical sciences (physics, chemistry, biology, etc.)
Code Of Ethics: I acknowledge that I and all co-authors of this work have read and commit to adhering to the ICLR Code of Ethics.
Submission Guidelines: I certify that this submission complies with the submission instructions as described on https://iclr.cc/Conferences/2025/AuthorGuide.
Reciprocal Reviewing: I understand the reciprocal reviewing requirement as described on https://iclr.cc/Conferences/2025/CallForPapers. If none of the authors are registered as a reviewer, it may result in a desk rejection at the discretion of the program chairs. To request an exception, please complete this form at https://forms.gle/Huojr6VjkFxiQsUp6.
Anonymous Url: I certify that there is no URL (e.g., github page) that could be used to find authors’ identity.
No Acknowledgement Section: I certify that there is no acknowledgement section in this submission for double blind review.
Submission Number: 7063
Loading